Brucellosis is a serious transboundary global zoonosis that is spread by livestock and wildlife and is primarily due to infection with Brucella abortus, B. melitensis or B. suis. Although the disease is serious in humans, human to human transmission is rare and nearly all cases of disease are due to contact with infected animals or animal products (Franco et al. (2007) The Lancet Infectious Diseases 7, 775-786). Brucellosis (with the exception of canine brucellosis) is caused by infection with smooth strains of Brucella, those that present O-polysaccharide (OPS) as part of the smooth lipopolysaccharide (sLPS) on their outer membrane (Cardoso et al. (2006) Microbial Cell Factories 5, 13).
The sLPS is a macromolecule comprising lipid A with a diaminoglucose backbone attached to a set of core sugars which, in the case of smooth Brucella strains, is attached to the OPS which extends into the extracellular environment. This OPS is a recognised and established virulence factor (Porte et al. (2003) Infection and Immunity 71, 1481-1490) and induces a strong antibody response in the host against which classical and contemporary serodiagnostic assays are directed (Nielsen & Yu (2010) Prilozi 31, 65-89). Unfortunately, infection with other Gram negative bacteria having OPS of similar structure, but which may be phylogenetically unrelated, may give rise to antibodies which cross react in the classical and contemporary serodiagnostic assays (Corbel (1985) Vet. Bull. 55, 927-942). Such false positive serological reactions (FPSRs) become particularly problematic in regions of low to zero prevalence of brucellosis, whereby the positive predictive value of the serological assays becomes extremely small.
Attempts to improve the specificity of brucellosis serodiagnosis have revolved mainly around efforts to identify and apply recombinant protein antigens. Whilst many studies have been published extolling the virtues of one antigen or another for the resolution of false positives, none have made it into assays that have progressed beyond the research laboratory into routine and effective use.
Antibodies raised owing to infection with Yersinia enterocolitica O:9 cross react significantly with the OPS from Brucella smooth strains, owing to the considerable similarity in the structure (McGiven et al. (2008) Journal of Immunological Methods 337, 7-15) and are believed to cause many of the false positive serological reactions in diagnostic testing (Gerbier et al. (1997) Vet Res 28, 375-383). Likewise, antibodies raised against the OPS from smooth Brucella strains show considerable reaction against Y. enterocolitica O:9 OPS (Jungersen et al. (2006) Epidemiol Infect 134, 347-357).
The Brucella OPS is an unbranched variable length homopolymer formed by, on average, 50-100 units of the rare sugar 4,6-dideoxy-4-formamido-D-mannopyranose (N-formyl perosamine) (Meikle et al. (1989) Infect Immun. 57, 2820-2828). The 4,6-dideoxy-4-formamido-D-mannopyranose unit itself is only found in nature within the OPS structure of Brucella organisms and of Y. enterocolitica O:9. In almost all Brucella organisms, neighbouring 4,6-dideoxy-4-formamido-D-mannopyranose units are variably α-1,2 or α-1,3 linked. The structural data demonstrates that it is only the α-1,3 linkage which differentiates Brucella and Y. enterocolitica O:9 OPS, as the latter is an exclusively α-1,2 linked 4,6-dideoxy-4-formamido-D-mannopyranose homopolymer (Caroff et al. (1984) Eur. J. Biochem. 139, 195-200). Between Brucella strains, the proportion of α-1,3 linkages varies between 2-20% (Meikle et al. (1989) Infect. Immun. 57, 2820-2828).
Even in Brucella strains with as much as 20% α-1,3 linkages, the OPS will contain many contiguous α-1,2 linkages (Bundle et al. (1987) Biochemistry 26, 8717-8726). It is not known, however, if the sequence of linkages is ordered and regular. There are three main biosynthesis pathways for OPS, the Wzx/Wzy, ABC (Wzm/Wzt) and Synthase systems (Wang et al. (2010) In “Endotoxins: structure, function & recognition”; Springer Netherlands, pp. 123-152). Brucella utilises the ABC system (Gonzalez et al. (2008) PLoS ONE 3, e2760), as is more commonly found in organisms that have homopolymeric OPS (including Y. enterocolitica O:9). In this pathway the saccharides are added to the growing OPS chain individually rather than being constructed into repeating units and then added, as is the case in the Wzx/Wzy pathway where OPS is frequently made up of repeating blocks of three to five different monosaccharides. As such, it is quite possible that there is no regularity to the location of the α-1,3 links and they are distributed stochastically with regions of extended contiguous α-1,2 linkages even in strains with a relatively high proportion of α-1,3 linkages. The control and fidelity of the enzymes that synthesise the OPS have yet to be established.
The proportion of the two linkage types affects the shape of the OPS and the different shapes are recognisable by monoclonal (Douglas et al. (1988) J Clin Microbiol 26, 1353-1356) and polyclonal antibodies such as those used for serotyping Brucella into A (low number of α-1,3 linkages, typically 2%) and M (relatively high proportion of α-1,3 linkages, typically 20%) dominant strains or serotypes (Alton et al. (1994) Techniques for the Brucellosis Laboratory, pages 53-54; INRA Editions, ISBN-10: 2738000428).
The B. suis biovar 2 strain (Zaccheus et al. (2013) PLoS One 8, e53941) and B. inopinata 1302 (Wattam et al. (2012) mBio 3: 00246-12) do not appear to contain any α-1,3 linkages. B. inopinata BO2 is a highly unusual and distantly related member of the Brucella genus as well as being extremely rare. Although it is a smooth strain, it is lacking most of the genes required for the synthesis of 4,6-dideoxy-4-formamido-D-mannopyranose OPS (Wattam et al (2012) mBio, 3:5). Its OPS structure is of a different form which has not yet been identified.
The relative proportions and distribution of α-1,2 and α-1,3 linkages within the Brucella and Y. enterocolitica O:9 homopolymeric OPS create distinct, but not necessarily completely described, antibody binding epitopes. In Brucella, there are three different antigenic epitopes which can be found in the OPS for which there has been firm structural evidence (Bundle et al. (1989) Infect. Immun. 57, 2829-2836) as summarised in Table 1:
TABLE 1OPS epitopesNameofNumber ofPresent inepitopeperosaminesCharacteristicswhich OPSC/Y3 to 4N-formyl perosamines areAll smooth Brucellaexclusively joined by α1,2strains and alsolinkagesY. enterocoliticaO:9A5 or moreN-formyl perosamines arePredominantlyjoined by α1,2 linkageswithin all A-dominant Brucellastrains and alsoY. enterocoliticaO:9M2-6At least one α1,3 linkPredominantlypresent with at least onewithin M-dominantadjacent α1,2 linkages;OPS Brucellalocation of α1,3 linkstrains but also,within epitope undefinedto a lesser extent,A-dominant strains.Not found inY. enterocoliticaO:9
Exclusively α-1,2 linked tri- and tetrasaccharide sequences are found in high abundance within the OPS from B. abortus, melitensis and suis as well as in the OPS from Y. enterocolitica O:9. Such sequences are termed ‘C/Y epitopes’ as they are common within all smooth strains of economically significant Brucella and also to Y. enterocolitica O:9. Monoclonal antibodies that bind such sequences are termed anti-C/Y.
Longer sequences of more than four saccharides that are exclusively α-1,2 linked are more likely to be found in Brucella strains with lower proportions of α-1,3 links. This also includes Y. enterocolitica O:9 which contains only α-1,2 links. Such sequences are termed ‘A epitopes’ and, of course, contain C/Y epitopes within them (C/Y epitopes being α-1,2 linked perosamine chains up to 4 saccharides in length, as outlined above). Monoclonal antibodies that bind such epitopes are termed anti-A antibodies. Strains of Brucella with OPS containing low proportions of α-1,3 links and, therefore, more abundant and longer sequences of uninterrupted α-1,2 links, are termed “A dominant” strains (including most strains of B. abortus), even though the OPS may contain a proportion of α-1,3 links at least once every 50 residues (Bundle et al. (1989) Infect. Immun. 57, 2829-2836). The OPS from A-dominant strains of course contains C/Y epitopes, as well as A epitopes.
An anti-C/Y antibody will be expected to bind to both a C/Y epitope and to an A-epitope, since the shorter C/Y epitope structure forms part of the longer A-epitope structure.
Sequences of saccharides that contain a single α-1,3 linkage with limited contiguous α-1,2 linkages are termed ‘M epitopes’. Monoclonal antibodies that bind to such sequences are termed anti-M antibodies. Strains of Brucella with a high proportion of α-1,3 links are termed M dominant strains. Such strains comprise C/Y epitopes, but fewer A epitopes, if any, since the series of α-1,2 linkages is “broken up” by the presence of more frequent α-1,3 linkages.
The structure of the M epitope has not been defined and, in detailed terms, may vary according to the antibody. However, any M epitope must be sufficient in size to enable the binding of monoclonal antibodies raised against antigens containing the α-1,3 linkage and be sufficiently limited in α-1,2 linkages so as not to bind effectively to antibodies raised against antigens (or parts of antigens) that are exclusively α-1,2 linked. Thus, the presence of the α-1,3 linkage is critical, as is a limitation on the number of α-1,2 linkages. The structure of these antigens has been partially described previously (Bundle et al. (1989) Infect Immun 57, 2829-2836) although the allocation of Brucella strains into A and M types (and also mixed A and M) predates this considerably. This allocation is a fundamental part of the classical biotyping of Brucella strains (Nielsen et al. (2009) “Bovine brucellosis” In: Manual of Diagnostic Tests & Vaccines for Terrestrial Animals 2009; Office International Des Epizooties, Paris, pg 10-19). Recent work suggests that the M dominant OPS contains a repeating structural determinant that contains one α-1,3 link for every three α-1,2 links. (Kubler-Kielb & Vinogradov (2013) Carbohydr. Res. 378, 144-147).
The typing of strains into A, M or mixed A and M (where there is a proportion of α-1,3 linkages of between 2-20% (Meikle et al. (1989) Infect Immun 57, 2820-2828)) is performed using sera from hyperimmunised rabbits. The rabbits are inoculated with repeated doses of killed Brucella cells of either A or M dominant type until a high antibody titre has been obtained. The polyclonal sera is then absorbed with the heterologous cell type (e.g., sera from rabbits hyperimmunised with A dominant strains are absorbed with M dominant cells) to remove antibodies that cross react. After careful selection and empirical testing, the process leaves a population of polyclonal (now termed monospecific) antibodies that are more specific to the immunising type and can be used in agglutination assays to determine the A, M or mixed A and M status of an untyped strain (Alton et al. (1994) Techniques for the Brucellosis Laboratory, pages 53-54; INRA Editions, ISBN-10: 2738000428).
Presumptive diagnosis of Brucellosis depends on detection of antibodies to Brucella A, C/Y and M antigens and is confirmed by microbiological culture (Nielsen et al. (2009) “Bovine brucellosis” In: Manual of Diagnostic Tests & Vaccines for Terrestrial Animals 2009; Office International Des Epizooties, Paris, pg 3-7). The A, C/Y and M antigenic determinants are expressed simultaneously on the O-antigen polysaccharide domain of Brucella smooth lipopolysaccharides (s-LPS) and this sLPS is used to detect antibodies present in sera of animals or humans suspected of being infected. Unfortunately, Brucella is a virulent pathogen that must be grown under level 3 bio-containment. This makes the production of diagnostic O-antigens a demanding, specialised and costly task, with significant health risks. Furthermore, since A, C/Y and M epitopes each may be expressed on a single Brucella sLPS and since this antigen is resistant to common partial degradation methods, it has proved difficult to isolate pure A or M antigenic determinants.
In summary, the Brucella OPS is highly immunogenic and many antibodies are raised against it in infected animals. This OPS contains a number of overlapping antibody epitopes, some of which are unique to Brucella and some that are not. The existence of non-unique epitopes within the Brucella OPS compromises its use as a serodiagnostic antigen and is a major factor in the occurrence of false positive serological results. It has previously been reported, by some workers (Alonso-Urmeneta et al. (1998) Clinical and Diagnostic Laboratory Immunology 5, 749-754), that antibodies to common OPS epitopes dominate the humoral response in cases of animal brucellosis and that the epitopic structure of the LPS, be it A, M or mixed A and M dominance, is not relevant in serodiagnosis.
However, given the problems of cross-reactivity of antibodies raised against different strains of Brucella, as well as other organisms, there is a need to identify antigens and methods capable of discriminating between antibodies raised against a Brucella bacterium and those raised against other organisms.